Subtle Motion Detection Using Wi-Fi for Hand Rest Tremor in Parkinson's Disease

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Wearable Technology and Health Monitoring

Background:

• Parkinson's disease (PD) affects over 60s, with symptoms fluctuating daily, complicating clinical assessment and treatment.
• Current diagnostic and medication strategies are limited by short clinical sessions, highlighting the need for continuous home health monitoring.
• Wireless sensing technology offers a non-invasive approach for remote patient monitoring.

Purpose of the Study:

• To develop and validate a method for detecting, classifying, and quantifying daily movements and motor symptoms in Parkinson's disease patients.
• To leverage wireless sensing technology and deep learning for improved Parkinson's disease healthcare.
• To enable precise, continuous monitoring of motor symptoms outside of clinical settings.

Main Methods:

• Utilized wireless sensing technology to capture Channel State Information (CSI) from Wi-Fi signals in the presence of human movement.
• Transformed CSI data into images for training a deep learning model.
• Employed modified VGG19 and Resnet152 deep learning architectures to distinguish daily movements and tremors.

Main Results:

• Achieved high accuracy in tremor recognition: 99.59% with modified VGG19 and 100% with modified Resnet152.
• Successfully segmented tremor movements from other daily activities.
• Quantified tremor characteristics, including frequency and duration.

Conclusions:

• The proposed wireless sensing and deep learning method is highly effective for detecting and quantifying Parkinson's disease tremors.
• This approach facilitates accurate, continuous home health monitoring for Parkinson's disease patients.
• The technology holds significant potential for improving the diagnosis, management, and overall healthcare of individuals with Parkinson's disease.